1 kishimoto - 日本言語学会Title 1_kishimoto.indd Created Date 4/27/2018 8:25:19 AM

言語研究（Gengo Kenkyu）153: 5–39（2018） doi: 10.11435/gengo.153.0_5

Projection of Negative Scope in Japanese

Hideki KishimotoKobe University

Abstract: On the basis of NPIs licensed under the scope of negation, this article shows that in Japanese, a negative head introduced in the head position of NegP extends its scope over TP if it undergoes Neg-head raising. While a grammati-cal negator (paired with lexical verbs) is subject to Neg-head raising, a lexical negator retaining original categorical properties as an adjective does not undergo Neg-head raising, in which case negative scope extends only over vP. By making use of the raising construction headed by the aspectual verb iru ‘be’, it is shown that subjects undergo raising to Spec-TP when the clause includes a nomina-tive argument, but that oblique subjects may remain in their base-generated predicate-internal position. It is also shown that there are two types of raising constructions formed on the verb naru ‘become’—one type in which the subject of the embedded verb is moved to the matrix clause, and the other in which the embedded subject moves only in the embedded clause.*

Key words: adjectival negator; negative scope; sika nai; raising construction; Japanese

1. IntroductionIn a language like Japanese with SOV word order, the putative results of movement operations, even if they apply, are often not visible in the surface strings. This gives rise to a number of controversies; for instance, in the Japanese literature, there has been a debate as to whether subjects are raised to Spec-TP by A-movement (i.e. subject raising) or remain in predicate-internal position (with no subject raising) (see e.g. Kuroda 1988, Miyagawa 1989a, 1989b, Fukui 1995, Kishimoto 2001). There is also an issue whether Japanese has instances of head movement—the most prominent issue being whether the verb remains within vP or is raised to T (see e.g. Otani and Whitman 1991, Sakai 1998, Fukui and Takano 1998, Koizumi 1999, 2000, Fukui and Sakai 2003). The main purpose of the present paper is to show that in Japanese, there are certain movement operations, including phrasal and head movement, that bring out some syntactic consequences, which shed light on the clause structure of the language. Specifically, it is shown that a nega-

* I am thankful to Hiroaki Tada, Mamoru Saito, Yuji Takano, Keiko Murasugi, Saeko Uru-shibara, Yu Yile, Hidekazu Tanaka, Yoshie Yamamori, and the participants of my graduate seminar at Kobe University for comments and suggestions on earlier versions of the pres-ent paper. I am also grateful to the two anonymous reviewers for helpful comments. Any remaining errors and inadequacies are my responsibility.

6 Hideki Kishimoto

tive head nai ‘not’ projecting syntactic negative scope is an element that undergoes head movement, and that the syntactic behavior of negative polarity items (NPIs) licensed under negative scope is affected by two types of movement—subject rais-ing and Neg-head raising.

In Japanese, negation is expressed by a clausal negator occupying a head posi-tion, and the most typical clausal negator is nai ‘not’. One notable fact of the nega-tive head nai is that the extent of negative scope changes in accordance with its head movement. A negative head introduced in the head position of NegP extends its scope over TP if it undergoes head movement. When no Neg-head raising takes place, the scope of negation extends only over vP (instead of TP), and a subject-object asymmetry is observed with regard to the licensing of NPIs. When negative scope falls over TP, it is not possible to discern whether subjects are located in Spec-TP or in predicate-internal position. Nevertheless, the raising con-struction (formed on the aspectual verb iru ‘be’), where long distance A-movement is induced, makes it possible to confirm that subjects undergo A-movement to TP when a nominative argument is included in the clause. Data on NPIs lead us to conclude that Japanese invokes both phrasal and head movement, even though their effects are not necessarily visible in the superficial sequences of words.

The present paper is organized as follows. First, section 2 discusses how nega-tive scope extends over TP in Japanese, and then, shows that while a non-adjectival (or decategoricalized) grammatical negator appearing with verbs is amenable to Neg-head raising, a clausal negator associated with adjectives retains the original categorical status of an adjective, and does not undergo Neg-head raising. On the basis of the aspectual construction headed by iru ‘be’, section 3 shows that subjects undergo raising to Spec-TP when the clause contains a nominative argument, and that oblique subjects may remain in their base-generated predicate-internal posi-tion. It is also shown that there are two types of naru-constructions—one type in which the nominative subject of the main verb is extracted from the embedded clause, and the other in which the raising of the nominative subject takes place only within the embedded clause. Section 4 presents a conclusion.

2. Negative scope and Neg-head raisingIn this section, on the basis of the behavior of negative polarity items (NPIs), which are licensed under the syntactically projected scope of negation, it is argued that the extent to which a negative head extends its scope differs depending on where it is located in clause structure.

2.1. Negative scope in simple verbal clausesLet us first go over the basic facts of NPIs, which allow us to assess the clause architecture of Japanese. Among several types of NPIs available in Japanese, NPIs formed with the particle sika ‘only’ are used, because sika can turn many types of DPs/PPs into NPIs by attaching to them.1 Broadly speaking, DPs with the par-

1 Kishimoto (2017) suggests that NPIs having the form wh-Q (e.g. dare-mo ‘anyone’ and

Projection of Negative Scope in Japanese 7

ticle sika, which serve as NPIs, are legitimate regardless of whether they appear in subject or object position, but cannot be included in affirmative clauses, as exem-plified in (1).

(1) a. Ken-sika hon-o {yoma-nakat-ta/*yon-da}. Ken-only book-acc {read-neg-past/read-past} ‘Only Ken {did not read/read} the book.’ b. Ken-ga kono hon-sika {yoma-nakat-ta/*yon-da}. Ken-nom this book-only {read-neg-past/read-past} ‘Ken {did not read/read} only this book.’

The examples in (2) illustrate that NPIs with sika need to be located within the finite clause in which a clausal negator appears.

(2) a. *Ken-sika [Mari-ga ko-naka-ta to] it-ta. Ken-only Mari-nom come-neg-past comp say-past ‘Only Ken said that Mari did not come.’ b. *Ken-ga [Mari-sika ki-ta to] iwa-nakat-ta. Ken-nom Mari-only come-past comp say-neg-past ‘Ken did not say that only Mari came.’

In (2a), the NPI Ken-sika appears in the matrix clause, but a negator is located in the embedded clause. Since the NPI is not c-commanded by the negator, the former falls outside the scope of the latter. Thus, (2a) is excluded as unacceptable. In (2b), on the other hand, the negator in the main clause c-commands the NPI in the embedded clause, but the sentence is excluded on the grounds that a finite clause boundary intervenes between the negator and the NPI (see e.g. Kato 1985).

There is one important difference that distinguishes Japanese from English: In English, unlike Japanese, a subject-object asymmetry is observed with regard to the licensing of NPIs, as illustrated in (3).

(3) a. *Anyone did not read the book. b. John did not read anything.

In both Japanese and English, a clausal negator looks like occurring in the same structural position, which is lower than TP but higher than vP. This suggests that the two languages have the layered clause structure: [TP [NegP [vP ]]]. Then, a ques-tion arises as to why NPIs are sanctioned in both subject and object positions in Japanese, as opposed to English.

In the literature on Japanese, there is the proposal that clause-wide negative

nani-mo ‘anything’) behave like floating quantifiers, appended to arguments. On the other hand, NPIs with sika behave as arguments, but not as floating quantifiers. For many speak-ers, the judgments on NPIs with sika are clear, but I occasionally came across speakers who are not really sensitive to the syntactic factors, and these speakers allow a DP with sika to be linked to a negator, regardless of their hierarchical position. The discussion in this paper is based on the judgments by the first group of speakers.

8 Hideki Kishimoto

scope is made available via Neg-head raising; i.e. Neg-head raising is responsible for the lack of a subject-object asymmetry in NPI licensing in Japanese. To be con-crete, on the assumption that a tense head is head-raised to Fin for the purpose of identifying the finiteness of the clause in Japanese (cf. Rizzi 1997, 2004, Radford 2009), Kishimoto (2013) suggests that negative scope extends over TP, which is occupied by the subject, as a consequence of Neg-head raising (see also Kishimoto 2007, 2008).

(4) [FinP [TP SUBJ [NegP [vP SUBJ OBJ V-v ] Neg ] Neg-T ] Neg-T-Fin]

�―――――――――�―――――――――� Negative Scope

In this analysis, when a negative head undergoes head raising, it first forms a complex head with T by virtue of Neg-head raising to T, and then, the entire head complex comprised of Neg and T is head-raised and occurs in the Fin-head posi-tion, as illustrated in (4).2 In this configuration, NPIs in subject position as well as NPIs in object position are licensed, because the complex head (including Neg) takes TP as its c-command domain. This analysis accounts for the difference in the behavior of NPIs between Japanese and English in terms of Neg-head raising. (The Neg-head raising analysis assumes that subjects are located in Spec-TP in both English and Japanese.) In Japanese, subjects in Spec-TP fall under the scope of negation, because negative nai is head-raised to Fin. In English, by contrast, the negative not does not undergo raising in clauses like (3), and hence its scope domain is limited to vP, located below NegP. Accordingly, English displays a subject-object asymmetry in the licensing of NPIs in (3).

Note that the predicate-internal subject hypothesis makes two subject posi-tions available—one is a predicate-internal subject position (Spec-vP (for verbal predicates) or Spec-aP (for adjectival predicates)), where a subject receives its theta role from the predicate, and the other, Spec-TP, which is the landing site of a subject when subject raising applies (see Kitagawa 1986, Kuroda 1988, Sportiche 1988, Koopman and Sportiche 1991, Fukui 1995, among others). Given this, another possibility suggests itself. In fact, there is another line of inquiry pursued in the literature, which attempts to account for the absence of a subject-object asymmetry in NPI licensing by assuming that subjects stay within vP (Aoyagi and Ishii 1994, Kato 1994, Watanabe 2004).

(5) [FinP [TP [NegP [vP SUBJ OBJ V-v] Neg ] T] Fin] �――――�――――� Negative Scope

According to the ‘subject-in-situ’ analysis, a negative head projecting scope is

2 In Japanese, the sentential negator is combined with tense to form a complex head. In English, and also in other European languages, sentential negators, even if they are realized as heads, function as elements independent of tense, and normally do not interact with it (see Haegeman 1995, Zanuttini 1997b, and others).


located in NegP without head raising in Japanese, as in (5). In Japanese, even though the negator takes scope only over vP, a subject NPI (as well as an object NPI) appears in a position low enough to be licensed by nai located in NegP. Thus, no subject-object asymmetry is observed in NPI licensing. On the other hand, in English, subjects are located in TP, falling outside the scope of negation, so that NPIs display a subject-object asymmetry, as observed in (3).

The second ‘subject-in-situ’ analysis might capture the Japanese facts of (1) without positing Neg-head raising, but it faces an empirical problem in account-ing for the well-formedness of (6), which includes the adjunct NPI kinoo-sika ‘only yesterday’.

(6) Kinoo-sika Ken-ga hataraka-nakat-ta. yesterday-only Ken-nom work-neg-past ‘Ken worked only yesterday.’

The temporal adverb kinoo ‘yesterday’ can appear only in clauses where the predi-cate appears in the past tense (see section 2.2). Given that modifiers are associated with the projections they modify (Radford 2009), it can be postulated that the temporal adjunct kinoo involves adjunction to TP.

In the second ‘subject-in-situ’ analysis, the negator appears in NegP, and the scope of negation extends only over vP. This leads to the prediction that the tem-poral adjunct kinoo-sika in (6) will not be licensed under the scope of negation, contrary to fact. On the other hand, the acceptability of (6) is naturally expected on the first Neg-head raising analysis, since the scope of negative nai extends over TP as a consequence of Neg-head raising to Fin. In light of this consideration, it is apparent that the Neg-head raising analysis is favored over the subject-in-situ analysis.

Under the Neg-head raising analysis, the clausal negator nai takes clause-wide scope when it undergoes Neg-head raising to T (and further to Fin). Obviously, in Japanese, this situation is obtained in verbal clauses, as shown in (1). The Neg-head raising analysis leads to the prediction that a subject-object (or subject-com-plement) asymmetry will be observed in the licensing of NPIs when Neg-head raising does not take place, because, in such cases, the scope of nai does not extend over TP. As I will discuss below, the kind of distribution predicted by the absence of Neg-head raising is found in non-verbal clauses (in Japanese).

2.2. Negative scope in non-verbal clausesBefore discussing how negative scope extends in non-verbal clauses, it is instruc-tive to consider the facts that provide an impetus for the Neg-head raising analysis. To be concrete, observe that in English, the aspectual verb have is subject to overt head raising, but its lexical counterparts are not (see Radford 1997).

(7) a. John has not finished his homework yet. b. John did not have his students take all the classes. c. John did not have his son examined by the doctor.

10 Hideki Kishimoto

The relative order of have and not differs between (7a) and (7b–c). In English, main verbs are not subject to head raising to T, so that the lexical verb have appears to the right of not, as in (7b) and (7c). In (7a), the aspectual have appears to the left of not as a consequence of head raising to T. Importantly, the aspectual verb have is a grammatical (or functional) word which does not possess lexical properties as a main verb, and thus, it is subject to head raising.

The Neg-head raising analysis, proposed by Kishimoto (2007, 2008), main-tains that an analogical pattern obtains in the head raising of negative nai in Japanese. Negative nai conjugates just like an ordinary adjective, which suggests that it probably originated as a lexical adjective. However, as I will see below, negative nai paired with verbal predicates now functions as a decategoricalized (i.e. deadjectival) grammatical negator (by virtue of losing its original lexical properties as an adjective). Thus, this type of nai is amenable to Neg-head raising.3 The nega-tive nai retaining its categorical status as an adjective can also be found. Generally, the negator nai associated with adjectives (including nominal adjectives) show properties that are expected if it retains the categorical status of an adjective, i.e. it behaves as a category that does not undergo head raising. (The negative nai proj-ects negative scope irrespective of whether it is categorized as a deadjectival nega-tor or an adjectival negator, because it serves to reverse the polarity of a proposi-tion.) It is reasonable to hypothesize here that the negative head nai undergoing head raising is a deadjectival grammatical marker, but that the negative nai occur-ring with adjectives is an adjectival negator that retains the categorical status as an adjective, so that it is not subject to head raising.

To substantiate the present claim, let us check whether negative nai associated with adjectives possesses the categorical status of an adjective, which differs from nai occurring with verbs. Whether or not a negative head has the categorical status of an adjective can be determined by way of embedding the negative clause under omou ‘think’, as discussed by Kishimoto (2007, 2008). The examples in (8) show that when omou takes a small-clause complement, adjectives (as well as nominal adjectives), but not verbs, are allowed to occur as the small-clause predicate.

(8) a. Eri-wa [sore-o omosiroku/hituyoo-ni] omo-u (koto-ga ar-u). Eri-top it-acc interesting/necessary think-pres fact-nom be-pres ‘(There are times when) Eri considers it interesting/necessary.’ b. *Ken-wa [hon-o ure-naku] omo-u (koto-ga ar-u). Ken-top book-acc sell.can-neg think-pres fact-nom be-pres (Lit.) ‘(There are times when) Ken considers books not to sell well.’

The embedded predicate immediately preceding omou ‘think’ is required to have adjectival inflection. The negated verb in (8b) qualifies as a predicate standing in front of the verb omou, at least, morphologically, owing to the adjectivally-inflect-

3 There are other factors that affect the possibility of Neg-head raising, but for the pres-ent purposes, it suffices to note that the categorical status of nai is a factor for determining whether its head raising takes place (see Kishimoto 2007, 2008).


ing nai. Nevertheless, (8b) is excluded, because nai here serves as a grammatical negator that does not retain the categorical status of an adjective. In contrast, negated adjectives can be placed in a small-clause complement to the verb omou ‘think’, as shown in (9).

(9) Ken-wa [sore-o omosiroku naku] omo-u (koto-ga ar-u). Ken-top it-acc interesting neg think-pres fact-nom be-pres ‘(There are times when) Ken considers it uninteresting.’

(9) might be a little awkward when it stands alone, but it is acceptable. The judg-ment is facilitated when the sentence is followed by a phrase like koto-ga aru ‘there are times when…’, while carrying the implication that the described situation holds rather exceptionally.4 In (9), the selectional requirement imposed on the small-clause predicate is not violated by nai occurring with the adjectives, which suggests that the negator has the categorical status of an adjective.

From the present perspective, if negative nai associated with adjectival predi-cates is a lexical adjective, it does not undergo Neg-head raising, i.e. nai remains in NegP where it is base-generated. This leads to the prediction that in adjectival clauses, the negative nai will take scope over aP, but not TP. In point of fact, the examples in (10) show that its scope does not extend beyond aP (located below NegP).

(10) a. *Asita-wa watasi-sika isogasiku na-i. tomorrow-top I-only busy neg-pres ‘Only I will be busy tomorrow.’ b. Watasi-wa {*asita-sika/sukosi-sika} isogasiku na-i. I-top {tomorrow-only/slightly-only} busy neg-pres ‘I will be busy {only tomorrow/only a little}.’

As shown in (10), where an adjectival predicate is negated, neither the subject NPI watasi-sika nor the temporal adjunct NPI asita-sika is sanctioned. On the other hand, the NPI sukosi-sika functioning as a predicate modifier to the adjective (which we can assume is adjoined to aP) is licensed. In this connection, note that a temporal adverb like asita specifying a specific point of time is sensitive to the tense form of the predicate, as (11) shows.

(11) Eri-wa asita {*hatarai-ta/hatarak-u/*isogasikat-ta/isogasi-i}. Eri-top tomorrow {work-past/work-pres/busy-past/busy-pres} ‘Eri {worked/will work/was busy/will be busy} tomorrow.’

4 One reviewer remarks that the small-clause construction embedding full-blown adjectives (including some negative adjectives such as moosiwake-nai ‘sorry’ and otonage-nai ‘childish’) is fully acceptable, but that (9) is awkward even if koto-ga aru follows it. On the other hand, some speakers I consulted find the same sentence in (9) fully acceptable even without koto-ga aru. It is not clear why speaker variation is found in regard to the judgments on this particular example.

12 Hideki Kishimoto

The fact suggests that asita is related to TP, which carries the tense information of the clause. Since, as noted earlier, modifiers are adjoined to the projections they modify (Radford 2009), it is easy to see that a temporal adverb like asita is adjoined to TP.5

Given that subjects as well as temporal adverbs like asita referring to a specific time frame fall outside the scope domain of nai, the configuration (12) can be assigned to the adjectival clauses in (10).

(12) [FinP [TP Advtemp SUBJ [NegP [aP SUBJ Pred-Mod A-a] Neg ] T] T-Fin] �――――�――――� Negative Scope

In (12), nai associated with the intransitive adjective isogasii ‘busy’ remains in NegP (without head raising). The negator does not extend its scope over TP, but takes scope over aP located below NegP. Consequently, the negator nai does not license the NPIs watasi-sika and asita-sika located in TP, while the predicate mod-ifier sukosi-sika is licensed. The fact suggests that nominative subjects are raised to Spec-TP in the clause headed by an adjectival predicate.6

Japanese has adverbials like syuu-hutuka ‘two days a week’ and nen-ni mikka ‘three days a year’, which specify frequencies (or intervals) rather than refer to a specific point of time (related to tense). A frequency adverb like syuu-hutuka behaves differently from asita ‘tomorrow’, although they both refer to a temporal relation in one way or another. In the first place, the occurrence of syuu-hutuka is not constrained by the tense of the predicate.

(13) Ano hito-wa syuu-hutuka {hatarai-ta/hatarak-u}. that person-top week-two.days {work-past/work-pres} ‘That person {worked/works} two days a week.’

Secondly, the two adverbs asita and syuu-hutuka show different behaviors in pseudo-cleft constructions where vP is focused.

(14) a. Ken-ga su-ru no wa [nikki-o {syuu-hutuka/?*asita} Ken-nom do-pres that top diary-acc {week-two.days/tomorrow} kaku] koto da. write fact cop ‘What Ken does is keep a diary {two days a week/tomorrow}.’ b. Ken-ga {syuu-hutuka/asita} su-ru no wa [nikki-o Ken-nom {week-two.days/tomorrow} do-pres that top diary-acc

kaku] koto da. write fact cop ‘What Ken will do {two days a week/tomorrow} is keep a diary.’

5 Tense information is encoded in the tense head that projects to TP. Thus, a temporal ad-junct is adjoined to TP, although the tense head is raised to Fin for its licensing.6 If nominative and accusative argumnets occur with sika, their morphological case marking is not overtly manifested, even though they carry structural Case features.


As discussed by Kishimoto (2016b), in the pseudo-cleft construction where vP is placed in its focus position, an adverb adjoined to vP can appear either in the ante-cedent clause or in focus position, but a TP adverb can appear only in the anteced-ent clause.7 The data in (14) suggest that syuu-hutuka is adjoined to vP, whereas asita is adjoined to TP. Furthermore, the NPI syuu-hutuka-sika, unlike the NPI asita-sika, is licensed in an adjectival clause.

(15) Ken-wa {*asita-sika/syuu-hutuka-sika} suugaku-no Ken-top {tomorrow-only/week-two.days-only} math-gen yosyuu-ga hituyoo-de na-i. preparation-nom necessary neg-pres ‘Ken needs to prepare for the math class {only tomorrow/only two days a week}.’

The facts of the NPI adverbs in (15) are expected if the frequency adverb syuu-hutuka, unlike the temporal adverb asita, is adjoined to aP.

In verbal clauses, the scope of negation extends over TP, due to the presence of Neg-head raising. Accordingly, both subject and temporal adverb NPIs are licensed by nai in (16).

(16) a. Kinoo-sika Ken-ga hasira-nakat-ta. yesterday-only Ken-nom run-neg-past ‘Ken ran only yesterday.’ b. Kinoo-wa Ken-sika hasira-nakat-ta. yesterday-top Ken-nom run-neg-past ‘Only Ken ran yesterday.’

Since the temporal adverb kinoo accompanying sika is adjoined to TP, it can be stated that the negative nai occurring with the verb takes scope over TP. The facts of NPI licensing in (16) follow if nai is raised to FinP, as illustrated in (17).

(17) [FinP [TP Advtemp SUBJ [NegP [vP SUBJ V-v] Neg ] Neg-T]Neg-T-Fin] �―――――――――�―――――――――� Negative Scope

The negative head nai appearing in verbal clauses is a decategoricalized gram-matical marker. The subject Ken-sika and the temporal adverb kinoo-sika are both legitimate in (16), because the negative head is raised to FinP and takes scope over TP. In section 3, it will be argued that when T carries the uninterpretable Case feature [+Nom], which values the unvalued Case feature on a nominative argu-ment, an EPP feature to motivate subject raising is assigned to it.8 Note that the

7 For some other restrictions imposed in vP-pseudo-cleft constructions, see Kishimoto (2016b).8 In the Case-assignment system, a Case-head assign Case to arguments (Chomsky 1981). Under the Agree system, Case features on arguments are valued by their probes. Pesetsky and Torrego (2001) suggest that nominative Case is an unvalued ‘tense’ feature on D, so that it is deleted in association with a tense feature on T. For Chomsky (2000, 2001), manifesta-

14 Hideki Kishimoto

adjective isogasii ‘busy’ in (10) selects a nominative subject, suggesting that the sub-ject should be raised to TP. In the examples in (10), negative scope does not extend over TP, and thus, the syntactic behavior of NPIs in (10) does indicate that subject raising is implemented in the adjectival clauses.

Recently, Shibata (2013, 2016) proposes an analysis of negative structure tak-ing the negator nai to occupy the same structural position as not in English, as in [TP [NegP [vP ] na]i]. His proposal is similar to the subject-in-situ analysis discussed in section 2.1, in the sense that the absence of a subject-object asymmetry in NPI licensing is attributed to the NPI subject (as well as the object NPI) staying below vP, over which nai takes scope. The example in (16a) poses an empirical problem on his analysis, however, because the temporal adverb kinoo, which is located above NegP, is licensed under the scope of negation.

If nai associated with adjectives does not undergo Neg-head raising, as pro-posed above, it is further predicted that transitive adjectives will display a subject-complement asymmetry in regard to the licensing of NPIs with sika. This predic-tion is borne out.

(18) a. *Kono kurasu-de-wa Ken-sika ano onnanoko-ni yasasiku this class-in-top Ken-only that girl-dat kind nakat-ta. neg-past ‘Only Ken was kind to that girls in this class.’ b. Kono kurasu-de-wa Ken-ga ano onnanoko-ni-sika yasasiku this class-in-top Ken-nom that girl-dat-only kind nakat-ta. neg-past ‘Ken was kind to only that girl in this class.’

Transitive adjectives like yasasii ‘kind’ and kibisii ‘strict’ take nominative subjects and dative complements. In (18), the dative-marked NPI with sika is licensed, but the nominative-marked NPI with sika is not.

A transitive adjective like hituyoo-da ‘necessary’ takes a dative-nominative rather than a nominative-dative case-marking pattern. With this type of predicate, the distributional pattern of NPIs is reversed, as shown in (19).

tion of structural Case depends on the probe; T values the Case feature on an argument as nominative, and v as accusative (see also Chomsky 2008, Radford 2009). Kishimoto (2017) proposes that a formal feature on the probe determines the Case value of an argument, and that finite T has [+Nom] and transitive v has [+Acc]. The present paper adopts the as-sumption that formal features on the probe (and not the probe itself ) Agree with the Case features of arguments, for this analysis has the advantage of providing an account for some case-marking constraints, e.g. the nominative-case constraint in Japanese, as discussed by Kishimoto (2017). I assume that this Case valuation system is not a matter of parametric variation. Recently, proposals are advanced in the literature to dispense with the EPP in fa-vor of the labeling algorithm (Chomsky 2013, 2015). I will leave open the theoretical ques-tion of how the EPP effects can be captured along this line of inquiry.


(19) a. *Ken-ni-sika ano okane-ga hituyoo-de nakat-ta. Ken-dat-only that money-nom necessary neg-past ‘Only Ken needed that money.’ b. Ken-ni-wa ano okane-sika hituyoo-de nakat-ta. Ken-dat-top that money-only necessary neg-past ‘Ken needed only that money.’

In (19), the nominative argument accompanying sika is licensed, but the dative argument with sika is not.

Obviously, the difference in the behavioral pattern of NPIs between (18) and (19) is correlated with the question of which argument serves as the subject of the clause. This can be confirmed by considering how subject-oriented zibun (which takes a subject located in aP or vP as its antecedent) behaves in the two types of transitive adjective clauses (Shibatani 1978, Kishimoto 2005).9(20) a. Keni-ga zibuni-no kodomo-ni-dake yasasikat-ta. Ken-nom self-gen child-dat-only kind-past ‘Ken was kind only to his own children.’ b. Keni-ni(-wa) zibuni-no okane-ga hituyoo-dat-ta. Ken-dat-top self-gen money-nom necessary-past ‘Ken needed his own money.’

With a transitive adjective like yasasii ‘kind’, subject-oriented zibun can take only the nominative argument as its antecedent. On the other hand, with hituyoo-da ‘necessary’, only the dative argument can be the antecedent of zibun. This shows that the subject of yasasii is the nominative argument, while the subject of hituyoo-da is the dative argument, and that these arguments fall outside the scope of nega-tion in (18a) and (19a).

The failure of licensing the NPI subjects in (18a) and (19a) comes from the fact that they are raised to TP, while the scope of negation does not extend over TP, due to the absence of Neg-head raising, as illustrated in (21).

(21) [FinP [TP SUBJ [NegP [aP SUBJ Compl A-a] Neg ]T]T-Fin] �―――�―――� Negative Scope

In adjectival clauses, subjects are located in Spec-TP, and thus, the nominative NPI subject in (18a) and the dative NPI subject in (19a), formed by attaching sika, are

9 In the causative construction (i), the embedded clause does not include tense (see Saito 2009), but the causee Mari (as well as the causer Ken) can be the antecedent of zibun.

(i) Keni-ga [Marij-o zibuni/j-no heya-de hatarak]-ase-ta. Ken-nom Mari-acc self-gen room-at work-caus-past (Lit.) ‘Ken made Mari work at self ’s room.’

This fact suggests that reflexivization is an operation targeting a subject located in Spec-vP (or its copy left there by movement).

16 Hideki Kishimoto

not licensed. By contrast, the NPI complements accompanying the particle sika in (18b) and (19b) are located below NegP, and hence are licensed by the negator nai, regardless of whether they appear in the nominative case or dative case.

The facts of NPI adjuncts lend further support to the present view. In both examples in (22), the temporal adjunct kinoo is prevented from accommodating the particle sika, but it is possible to append sika to the predicate modifier sukosi.

(22) a. Ken-wa {*kinoo-sika/sukosi-sika} kodomo-ni yasasiku nakat-ta. Ken-top {yesterday-only/little-only} child-dat kind neg-past ‘Ken was kind to the children {only yesterday/only a little}.’ b. Ken-ni-wa {*kinoo-sika/sukosi-sika} okane-ga hituyoo-de Ken-dat-top {yesterday-only/little-only} money-nom necessary nakat-ta. neg-past ‘Ken needed money {only yesterday/only a little}.’

The predicate modifier sukosi appears in a position lower than NegP, while kinoo is adjoined to TP. Thus, the data in (22) suggest that the scope of nai appearing in adjectival clauses is limited to aP, projected below NegP, as illustrated in (21).

Incidentally, the negative nai associated with verbal predicates has morpho-logical status different from the negative nai occurring with adjectival predicates, in the sense that in the former, but not the latter, supportive verb insertion is implemented when a focus particle like mo ‘also’ intervenes between the predicate and nai, as in (23).

(23) a. Ken-ga {hasira-naka-ta/hasiri-mo si-naka-ta}. Ken-nom {run-neg-past/run-also do-neg-past} ‘Ken did not (even) run.’ b. Sore-wa utukusiku(-mo) nakat-ta. that-top beautiful(-also) neg-past ‘That was not (even) beautiful.’

The negator appearing with the verb hasiru ‘run’ is a bound morpheme morpho-logically, but the negator paired with the adjective utukusii ‘beautiful’ is not. One might claim then that nai associated with verbs undergoes Neg-head raising because it is a bound element.10 On the contrary, the difference in the morpho-logical status of nai does not tightly correlate with the distinction of ‘raising’ versus ‘non-raising’ Neg-heads. This can be verified by the syntactic behavior of the nega-tive nai occurring with iru ‘need’. First, observe an asymmetry in the licensing of NPI arguments in (24).

10 Another related issue concerns the form of negative nai. When a verb is negated, the negative nai appears in the mizen ‘irrealis’ form. But when an adjective is negated, nai appears in the renyoo ‘adverbial’ form. Needless to say, the difference in Neg-form is not tightly correlated with the possibility of Neg-raising, either.


(24) a. *Ken-ni-sika suugaku-no yosyuu-ga ira-na-i. Ken-dat-only math-gen preparation-nom need-neg-pres ‘Only Ken needs to prepare for his math class.’ b. Ken-ni-wa suugaku-no yosyuu-sika ira-na-i. Ken-dat-top math-gen preparation-only need-neg-pres ‘Ken needs to prepare for only his math class.’In (24), the nominative object of iru ‘need’ with sika is licensed under the scope of negation, but the dative subject with sika is not. Further, this negated predicate gives rise to a difference in acceptability with regard to the licensing of adjunct NPIs as well.(25) Watasi-ni-wa {*kyoo-sika/sukosi-sika} okane-ga ira-na-i. I-dat-top {today-only/little-only} money-nom need-neg-pres ‘I need money {only today/only a little}.’In (25), the predicate modifier sukosi-sika is licensed under the scope of negation, but the temporal adverb kyoo-sika is not. The failure of the negative nai to license the temporal adverb kyoo-sika and the dative subject Ken-ni-sika illustrates that the scope of negation does not extend over TP. This fact shows that the negative nai appearing with iru ‘need’ does not undergo Neg-head raising. Crucially, morpho-logical support is necessary for the negative nai associated with iru ‘need’ when an adverbial particle intervenes between them.

(26) Ken-ni-wa okane-ga mattaku {ira-nakat-ta/iri-mo Ken-dat-top money-nom at.all {need-neg-past/need-also si-nakat-ta}. do-neg-past} ‘Ken does not {need/even need} money at all.’

With iru ‘need’, just like ordinary verbs, when nai is separated from the host predi-cate, the supportive verb suru ‘do’ is inserted to its left, as in (26), thus showing that negative nai associated with this predicate is a bound morpheme (in morphologi-cal terms). Nevertheless, this negative element is not subject to Neg-head raising.

The facts regarding the behavior of NPIs in (24) and (25) plus the morpholog-ical property of nai shown in (26) suggest that whether or not nai undergoes Neg-head raising is not determined by its morphological status. Rather, the distinction is correlated with its categorical status. With a view to providing confirmation on this claim, let us consider whether ira-nai ‘need not’ can appear in the small-clause complement selected by omou ‘think’.(27) Ken-wa [kono hon-o ira-naku] omo-u (koto-ga ar-u). Ken-top this book-acc need-neg think-pres fact-nom be-pres ‘(There are times when) Ken considers this book unnecessary.’The sentence in (27), where the negated verb ira-nai ‘not need’ occurs as a small-clause predicate, is acceptable. This fact suggests that the negator nai appearing with iru ‘need’ functions as an adjective in categorical terms. Note, further, that the

18 Hideki Kishimoto

predicate iru ‘need’ looks like a verb, but does not have the categorical status as a verb, as confirmed by the predicate’s inability to appear in the subordinate clause selected by hosii ‘want’.

(28) Watasi-wa [Ken-ni sore-ga {deki-te/*it-te}] hosi-i. I-top Ken-dat it-nom {can.do-ger/need-ger} want-pres ‘I want Ken to {be able to do/need} it.’

As discussed by Kishimoto (2008), the desiderative predicate hosii ‘want’ allows a lexical verb to appear in the embedded clause. Nevertheless, the predicate iru ‘need’ cannot appear in the embedded clause, despite the fact that it conjugates like a verb. The fact points to the conclusion that iru is not a verbal predicate (categori-cally). Perhaps, this can be taken as a symptom that the predicate iru ‘need’ has been decategoricalized through grammaticalization and has lost the status of a lexical verb. In any event, given that the non-adjectival negator nai is paired with lexical verbs, there is a sense in which the necessity predicate iru selects for an adjectival negator, which is not subject to Neg-head raising.

When negated, the necessity predicates hituyoo-da ‘necessary’ and iru ‘need’ take an adjectival negator, which does not undergo Neg-head raising, but other necessity predicates like hituyoo-to suru ‘need’ and hituyoo-ni naru ‘become neces-sary’ take a clausal negator amenable to Neg-head raising. Thus, with the latter predicates, temporal adjunct NPIs, as well as predicate modifier NPIs, are licensed by nai, as shown in (29).

(29) a. Ken-ga {kinoo-sika/sukosi-sika} okane-o hituyoo-to Ken-nom {yesterday-only/little-only} money-acc need si-nakat-ta. do-neg-past ‘Ken needed money {only yesterday/only a little}.’ b. Ken-ni-wa {kinoo-sika/sukosi-sika} okane-ga hituyoo-ni Ken-dat-top {yesterday-only/little-only} money-nom necessary nara-nakat-ta. become-neg-past ‘Ken needed money {only yesterday/only a little}.’

Furthermore, in both sentences in (30), headed by hituyoo-to suru and hituyoo-ni naru, the subject NPIs with sika are licensed under the scope of negation.

(30) a. Ken-sika okane-o hituyoo-to si-nakat-ta. Ken-only money-acc need do-neg-past ‘Only Ken needed the money.’ b. Ken-ni-sika okane-ga hituyoo-ni nara-nakat-ta. Ken-dat-only money-nom necessary become-neg-past ‘Only Ken needed the money.’

The data illustrate then that nai occurring with hituyoo-to suru and hituyoo-ni naru extends its scope over TP. In addition, observe that these predicates cannot be


embedded in the small-clause complement of omou ‘think’.

(31) Ken-wa [sore-o {*hituyoo-to si-naku/?*hituyoo-ni nara-naku}] Ken-top it-acc {need do-neg/necessary become-neg} omo-u (koto-ga ar-u). think-pres fact-nom be-pres ‘(There are times when) Ken considers it necessary.’

The example in (31) indicates that the negative nai appearing with hituyoo-to suru and hituyoo-ni naru is classified as a deadjectival negator that extends its scope over TP as a consequence of Neg-head raising. Apparently, this comes from the fact that the negator is anchored to the verbs suru ‘do’ and naru ‘become’.

To summarize, negative nai associated with adjectival predicates, as well as nai paired with the necessity predicate iru ‘need’, retains the categorical status of an adjective. The adjectival type of sentential negator nai does not undergo Neg-head raising and hence does not take scope over TP.11

3. Subject raising in complex clausesUnder the view held here, a decategoricalized grammatical marker nai (paired with verbs) takes clause-wide scope owing to its Neg-head raising. In simple ver-bal clauses, then, NPI subjects are licensed regardless of whether they are raised to Spec-TP or remain within vP. Accordingly, simple verbal clauses cannot be used for assessing whether subjects undergo raising. Nevertheless, it is possible to dis-

11 There are some exceptional cases where negative nai appearing with adjectives is subject to Neg-head raising. The adjectives suki-da ‘fond of ’ and hosii ‘want’ allow their associated negator to undergo Neg-head raising, so that in the clauses headed by these adjectives, negative scope extends over TP, as shown by (i).

(i) a. Ken-sika gohan-ga hosiku-nakat-ta. Ken-only meal-nom want-neg-past ‘Only Ken wants meals.’ b. Ken-ga gohan-sika hosiku-nakat-ta. Ken-nom meal-only want-neg-past ‘Ken wants only meals.’

When the negated predicate hosiku nai ‘not want’ is embedded under omou, the sentence is deviant.

(ii) ?*Ken-wa [gohan-o hosiku naku] omo-u (koto-ga ar-u). Ken-top meal-acc want neg think-pres fact-nom be-prs ‘(There are times when) Ken does not want meals.’

The deviance of (ii) shows that the negator associated with the adjectival predicate hosii is a functional negator. It is suggested by Shimizu (2013) that some emotional predicates, including suki-da and hosii, have been derived from their verbal counterparts. If this is the case, it is plausible to postulate that a deadjectival negator nai is associated with hosii and suki-da, because they somehow retain certain verbal properties (as their idiosyncrasies).

20 Hideki Kishimoto

cern whether subjects are located in Spec-TP or in predicate-internal position by looking at raising constructions which allow the subject of the embedded predicate to be moved from the embedded clause to the matrix clause.

Drawing on the raising construction headed by the aspectual verb iru ‘be’, section 3.1 shows that when tense (T) has a Case feature to license a nomina-tive argument, it also has an EPP feature to attract subjects, i.e. subject raising is invoked when the clause contains a nominative argument. In section 3.2, it is argued that in another type of raising construction formed on naru ‘become’, the possibility of subject raising out of the subordinate clause differs according to the type of complement clause it takes.

3.1. The aspectual constructionAs noted earlier, the predicate-internal subject hypothesis makes two subject posi-tions available in simple clauses. Logically, subjects could be located in a predicate-internal position (if no subject raising takes place) or in Spec-TP (if they undergo subject raising). In Japanese, there is an issue as to which position subjects occupy. For instance, Fukui (1995) and Kuroda (1988) hold that subjects appear in predi-cate-internal position without subject raising, while Miyagawa (1989b, 2001) and Kishimoto (2001) maintain that subjects are raised to Spec-TP by virtue of the EPP requirement imposed on T.(32) a. [TP [vP SUBJ [VP V ] v] T] b. [TP SUBJ [vP SUBJ [VP V ] v] T]The discussion of the structural position of subjects is often confined to cases involving nominative subjects, but more recently, a different claim has been advanced in Kishimoto (2010), to the effect that the structural position of subjects varies depending on their marking; that is, nominative subjects are raised to TP, whereas subjects marked with oblique kara ‘from’ or de ‘with’ remain within the predicate (cf. Ueda 2003).

In simple verbal clauses, the constituent position of subjects cannot be evalu-ated by way of the syntactic behavior of NPIs. The reason is that the same results follow regardless of whether subject NPIs undergo raising to Spec-TP or remain in vP. Nevertheless, it is possible to assess the constituent position of subjects by making use of one type of raising construction formed on the aspectual verb iru ‘be’.(33) Ken-ga hon-o yon-de i-ru. Ken-nom book-acc read-ger be-pres ‘Ken is reading the book.’One remarkable feature of the aspectual construction in (33) is that the clausal negator nai can either precede or follow the aspectual verb (but must follow the main verb).(34) a. Ken-ga hon-o yon-de i-na-i. Ken-nom book-acc read-ger be-neg-pres ‘Ken is not reading the book.’ V-BE-NOT


b. Ken-ga hon-o yoma-nai-de i-ru. Ken-nom book-acc read-neg-ger be-pres ‘Ken is not reading the book.’ V-NEG-BEIn particular, the type of aspectual construction given in (34b), where a negator is embedded under the aspectual verb iru ‘be’, allows us to assess whether a subject is raised to Spec-TP or not, because negative scope does not extend over TP in the matrix clause.

The aspectual verb iru takes a raising complement, as confirmed by the fact that inanimate subjects are allowed, as in (35a), as well as the fact that clausal idi-oms can be embedded with no loss of their idiomatic meanings, as in (35b) (see Carnie 2006).(35) a. Sora-ga mada {hare-nai-de i-ru/hare-te i-na-i}. sky-nom still {clear-neg-ger be-pres/clear-ger be-neg-pres} ‘The sky has not cleared yet.’ b. Kono mise-de-wa imadani kankodori-ga {naka-nai-de this store-at-top still cuckoo-nom {sing-neg-ger i-ru/nai-te i-na-i}. be-pres/sing-ger be-neg-pres} ‘There are still some customers at this store.’Note that inanimate subjects and clausal idioms are not allowed to appear in con-trol constructions, as exemplified in (36).(36) a. *Sora-ga hare-ta-i. sky-nom clear-want-pres ‘The sky wants to clear.’ b. *Kono mise-de-wa kankodori-ga naki-ta-i. this store-at-top cuckoo-nom sing-want-pres ‘The cuckoo wants to sing at this store.’The predicate -tai ‘want’ takes a control complement, as [Subj [PRO V]-tai]. Because PRO needs to refer to an animate entity (see Kishimoto 2005), the sen-tences in (36) are not acceptable. These sentences are ruled out, since the subject in (36a) is not animate, and the subject idiom in (36b) does not contain a referential subject. The data in (35) show that (33) is a raising construction whose subject is generated in the embedded clause, i.e. the upper verb iru does not impose selectional restrictions on the subject.12 The variant of the aspectual construction

12 In control constructions, neither inanimate subjects nor subject idioms can be embedded. This restriction is observed in the auxiliary verb constructions in (i).

(i) a. *Sora-ga hare-te oi-ta. sky-nom clear-ger put-past ‘The sky cleared.’ b. *Kono mise-de-wa mada kankodori-ga nai-te oi-ta. this store-at-top still cuckoo-nom sing-ger put-past ‘The cuckoo still sang at this store.’

22 Hideki Kishimoto

in (34b), where nai precedes the aspectual verb iru, has an agentive implication semantically, on the basis of which Takezawa (2004) argues that it has a control structure. Nevertheless, the pattern of distribution observed above suggests that (34b), as well as (34a), is a raising construction.

In the aspectual construction, the non-finite TP in the te-complement clause does not serve as the (final) landing site of the subject undergoing subject raising. Therefore, the configuration in (37a) can be posited for (33) if the thematic subject of the main verb undergoes subject raising. On the other hand, (33) has the struc-ture in (37b) if no subject raising is implemented.

(37) a. [FinP [TP SUBJ [vP [FinP [TP SUBJ [vP SUBJ [VP OBJ V]v]T]T-Fin]BE]T]T-Fin] b. [FinP [TP [vP [FinP [TP [vP SUBJ [VP OBJ V]v]T]T-Fin]BE]T]T-Fin]

In (37a), the matrix T, as well as the embedded T, is assigned an EPP feature, so that the nominative subject is moved to the matrix TP by way of the embedded TP (Chomsky 1995, Bošković 2002). The aspectual construction has a bi-clausal structure, and thus, the extent to which the scope of the negative nai extends dif-fers according to where it appears. In particular, when the negator precedes the aspectual verb, negative scope does not extend over the matrix TP, but is limited to the embedded TP, for the Neg-head resides in the embedded clause. On the basis of this aspectual construction, it can be confirmed that in Japanese, nominative subjects undergo raising.

Let us now look at how NPIs with sika behave in the aspectual construction. When nai intervenes between the main and the aspectual verbs, a difference in acceptability is observed with regard to the licensing of NPIs, as in (38).

(38) a. Gakusei-ga hon-sika yoma-nai-de i-ru. student-nom book-only read-neg-ger be-pres ‘The student has been reading only books.’ b. *Gakusei-sika hon-o yoma-nai-de i-ru. student-only book-acc read-neg-ger be-pres ‘Only the student has been reading books.’

While the NPI object in (38a) is licensed, the NPI subject in (38b) is not. A simi-lar pattern of distribution is found in (39), which involves adjunct NPIs.

(39) Ken-wa {*asita-sika/koko-de-sika} hataraka-nai-de i-ru. Ken-top {tomorrow-only/here-in-only} work-neg-ger be-pres ‘Ken has been working {only tomorrow/only here}.’

The locative adjunct koko-de specifies the place of the event described by the main verb, which suggests that it appears in the embedded clause. On the other hand, the choice of the temporal adverb asita is affected by the tense form of the matrix aspectual verb iru ‘be’.

In (i), the auxiliary verb oku ‘put’ selects a control complement, and since PRO needs to re-fer to an animate entity, the examples in (i) are excluded.


(40) Ken-wa asita hataraka-nai-de {i-ru/*i-ta}. Ken-top tomorrow work-neg-ger {be-pres/be-past} ‘Ken {will be/was} working tomorrow.’

This fact suggests that the adverb asita is adjoined to the matrix TP.13 Given that the subject in (38b) patterns with asita in (39) with regard to the licensing of NPIs with sika, it is fair to state that the negative head nai occurring between the main verb and the aspectual verb iru is raised to the embedded FinP, but not any further, whereas the nominative subject is moved to the matrix Spec-TP, as depicted in (41).

(41) [FinP [TP SUBJ [vP [FinP [TP [NegP [vP SUBJ OBJ V-v]Neg]T]NEG-T-Fin]Be]T]T-Fin] �――――�――――� Negative Scope

When the negative nai is embedded under the aspectual verb, the negative scope is extended only over the embedded clause. In (38), a subject-object asymmetry is observed in regard to NPI licensing, since the subject, but not the object, is extracted from the scope domain of nai.

In this connection, observe that a numeral quantifier with sika, which is floated from the subject, is licensed by nai in the embedded clause, as shown in (42).

(42) {Gakusei-ga hito-ri-sika/*Kono gakusei-sika} hasira-nai-de i-ru. {student-nom one-cl-only/this student-only} run-neg-ger be-pres ‘Only {one/this} student has been running.’

The contrast in acceptability in (42) suggests that while the subject is raised to the matrix TP, the numeral quantifier hito-ri ‘one person’ associated with the subject is allowed to occur in the embedded clause. This fact is naturally expected, because a numeral quantifier launched off the subject can be appended to its copy in vP-internal position (Miyagawa 1989).

In the aspectual construction, if nai is placed in the matrix clause, it takes scope over the matrix clause, and no subject-object asymmetry is found in NPI licensing,

13 In simple clauses, the choice of a temporal adjunct depends on the tense form of the predicate. In the aspectual construction formed on the aspectual verb iru ‘be’, this correla-tion is not necessarily obtained, as exemplified in (i).

(i) Ken-wa kinoo hasit-te {i-ru/i-ta/i-na-i/i-nakat-ta}. Ken-top yesterday run-ger {be-pres/be-past/be-neg-pres/be-neg-past} ‘Ken {is/was not/is not/was not} running yesterday.’

Example (i) shows that a temporal adverb referring to a past time can appear even when the aspectual verb takes the present form. This is because (i) can have an experiential interpreta-tion where the embedded clause is taken to describe a past event. This experiential interpre-tation is not available for the nai-de iru construction, so that kinoo ‘yesterday’ is allowed only when the aspectual verb appears in the past form. For obvious reasons, the experiential use is not available when the adverb asita ‘tomorrow’ referring to a future is used.

24 Hideki Kishimoto

as shown in (43).

(43) a. Saikin Ken-ga hon-sika yon-de i-na-i. recently Ken-nom book-only read-ger be-neg-pres ‘Recently, Ken has been reading only books.’ b. Saikin Ken-sika hon-o yon-de i-na-i. recently Ken-only book-acc read-ger be-neg-pres ‘Recently, only Ken has been reading books.’

No difference in acceptability is observed between the two types of NPI adjuncts with sika, i.e. kinoo-sika and koko-de-sika, either, if the negator follows the aspectual verb.

(44) a. Ken-wa kinoo-sika hatarai-te i-nakat-ta. Ken-top yesterday-only work-ger be-neg-past ‘Ken was working only yesterday.’ b. Zutto Ken-wa koko-de-sika hatarai-te i-nakat-ta. all.the.time Ken-top here-in-only work-ger be-neg-past ‘Ken was working only here all the time.’

The data suggest that in the aspectual construction where nai follows the aspectual verb iru, nai is raised to the matrix Fin, as illustrated in (45).

(45) [FinP [TP SUBJ [NegP [vP [FinP [TP [vP SUBJ OBJ V-v]T]T-Fin]Be]Neg]T]NEG-T-Fin] �――――――――――�――――――――――� Negative Scope

When the negative head nai appears in the matrix clause, its scope extends over the entire clause by virtue of its head movement. Accordingly, in (43), the subject and the object of the main verb are both licensed under the scope of negation.

As remarked earlier, the discussion on subject raising in Japanese is often con-fined to cases where subjects receive nominative case. But note that subjects can bear case markings other than nominative case (Inoue 1998, Kishimoto 2005).

(46) a. Ken-ni sono zi-ga mie-ta. Ken-dat that letter-nom see-past ‘Ken was able to see that letter.’ b. Watasi-kara sono koto-o hanasi-ta. I-from that matter-acc talk-past ‘I talked about that matter.’ c. Kodomo-tati-de atumat-ta. child-pl-with get.together-past ‘The children got together.’

The subject is marked with dative case in (46a). Dative subjects appear mainly in clauses headed by transitive stative predicates. In (46b), the subject bears the abla-tive kara ‘from’, since it is thematically identified as a source, as well as an agent (Kishimoto 2010). In (46c), the subject is assigned de ‘with’, for it is an agent argu-


ment referring to a group of people (Kishimoto 2005, Takubo 2010).The initial arguments bearing different markings in the three clauses in (46) all

behave as subjects syntactically. This is confirmed by the fact that they can be the antecedents of zibun, which takes a subject in vP as its antecedent.

(47) a. Keni-ni zibuni-no ie-ga mie-ta. Ken-dat self-gen house-nom see-past ‘Ken saw his own house.’ b. Keni-kara-wa zibuni-no koto-o hanasa-nakat-ta. Ken-from-top self-gen matter-acc speak-neg-past ‘Ken did not talk about his own matter.’ c. Kodomo-tatii-de zibuni-(tati)-no keikaku-o tate-ta. child-pl-with self-pl-gen plan-acc make-past ‘The children drew up their own plans.’

Subject honorification, which targets a subject located in vP, provides another type of corroboration for the adequacy of the present view (Harada 1976, Hasegawa 2006).14(48) a. Sensei-ni sore-ga o-mie-ni-nat-ta. teacher-dat it-nom hon-see-dat-become-past ‘The teacher saw it.’ b. Sensei-kara sono koto-o o-hanasi-ni-nat-ta. teacher-from that fact-acc hon-speak-dat-become-past ‘The teacher talked about that matter.’ c. Sensei-tati-de o-atumari-ni-nat-ta. teacher-pl-with hon-get.together-dat-become-past ‘The teachers got together.’

Given that the italicized arguments in (46) can be targeted for subject honorifica-tion, and can also serve as the antecedents of subject-oriented reflexive zubin, it can be stated that they serve as subjects, regardless of their marking.

Let us now turn to the question of how various types of subjects behave when they are embedded in the nai-de iru construction. To begin with, note that the dative-subject construction gives rise to a subject-object asymmetry in NPI licens-ing, as in (49).

(49) a. *Zutto Ken-ni-sika sonna undoo-ga deki-nai-de all.the.time Ken-dat-only such exercise-nom can.do-neg-ger

14 In (i), the embedded subject is targeted for subject honorification, even if it appears in the small clause, which does not comprise a tense element (Takezawa 1987).

(i) Watasi-wa [Sasaki-sensei-o totemo o-utukusiku] omo-u. I-top Sasaki-teacher-acc very hon-beautiful think-pres ‘I think Ms. Sasaki very beautiful.’

This fact suggests that subject honorification picks out an argument located in vP.

26 Hideki Kishimoto

i-ru. be-pres ‘Only Ken has been able to do such exercises all the time.’ b. Zutto Ken-ni-wa sonna undoo-sika deki-nai-de all.the.time Ken-dat-top such exercise-only can.do-neg-ger i-ru. be-pres ‘Ken has been able to do only such exercises all the time.’

The data show that in the dative-subject construction, the dative subject is raised to Spec-TP, whereas an object is not even if it is marked with nominative case.

Obliquely-marked NPI subjects with sika behave differently. In the aspectual construction where nai is located in the complement clause, the NPI subject marked with the oblique de ‘with’ or kara ‘from’ is licensed by nai, as in (50).

(50) a. Zutto kodomo-tati-de-sika ryokoo-no keikaku-o all.the.time child-pl-with-only trip-gen plan-acc tate-nai-de i-ru. make-neg-ger be-pres ‘Only the children have been planning on their trips all the time.’ b. Zutto hahaoya-kara-sika hanasi-o si-nai-de i-ru. all.the.time mother-from-only talk-acc do-neg-ger be-pres ‘Only the mother has been talking all the time.’

In (50), the negative scope does not extend over the matrix clause. The accept-ability of the sentences in (50) shows then that the oblique subjects remain in situ without raising to the matrix Spec-TP.

The data indicate that dative subjects, just like nominative subjects, undergo raising to Spec-TP, while oblique subjects do not. One question that arises at this point is why it is that the nominative and the dative subjects undergo raising to Spec-TP. In this connection, note that both nominative-subject and dative-subject constructions need to comprise a nominative argument (due to the nominative-case constraint: see Shibatani 1987).15 As is well-observed (Takezawa 1997, Koizumi 1999, and many others), the availability or unavailability of a nominative argument is correlated with the question of whether the clause has a finite tense.16 In the light of this fact, I suggest that when T carries the uninterpretable Case feature [+Nom] to license the Case feature of a nominative argument, an EPP fea-ture (to motivate subject raising) is imposed on T (cf. Pesetsky and Torrego 2001,

15 Shibatani (1978) claims that the nominative-case constraint is a general case-marking constraint in Japanese. Inoue (1998, 2007) observes that when subjects are marked with oblique case, the clause is exempt from the nominative-case constraint, but does not provide any explanation as to why this constraint does not apply in this case. See Kishimoto (2016a) for the discussion on this point.16 Kuno (1973) and Tada (1992) take the view that stative predicates license nominative case on their objects by virtue of their stativity.


Chomsky 2004, 2008).To be a little more concrete, in both nominative- and dative-subject con-

structions, a nominative argument needs to be included in the clause, which suggests that T should possess both nominative Case feature and EPP feature.17 Thus, in the nominative-subject construction, subject raising takes place, as (51) illustrates.18(51) [TP SUBJ[+NOM] [vP SUBJ[+NOM] V-v ] T [+NOM, EPP] ]

In (51), the Case feature on T is deleted after it values the Case feature on the nominative subject, and the EPP feature is deleted via the raising of the subject. The derivation of the nominative-subject construction in (51) converges when subject raising takes place.

In the dative-subject construction in (46a), T establishes an Agree relation with both dative subject and nominative object. I assume here, with Chomsky (2001), that the dative argument bears inherent Case with structural properties, so that T enters into an Agree relation with the dative subject (cf. Ura 1999).

(52) [TP SUBJ-dat [vP SUBJ-dat OBJ[+NOM] V-v ] T [+NOM, EPP] ]

In (52), the Case feature on T is deleted via Agree with the nominative object, and the dative subject, which is structurally closer to T than the nominative object, is raised to Spec-TP to satisfy the EPP requirement. Note that T does not value the Case feature of the dative argument, and that Case valuation is executed only between T and the nominative object. The derivation in (52) converges, since all the formal features can be deleted. In the oblique-subject constructions in (46b) and (46c), by contrast, no nominative argument appears in the clause, so T without any formal features is merged.

(53) [TP [vP SUBJ-abl/instr V-v] T ]

The inert type of T associated with the oblique-subject constructions does not include an EPP feature, so that no subject raising takes place, as (53) illustrates.

In regard to the derivation of the dative subject construction, one reviewer raises the question of why it is that the dative subject with sika is not licensed in (54), where the nominative object is scrambled to the sentence-initial position.

(54) *Zutto sonna undoo-ga Ken-ni-sika deki-nai-de all.the.time such exercise-nom Ken-dat-only can.do-neg-ger i-ru. be-pres ‘Only Ken has been able to do such exercises all the time.’

In the dative-subject construction, both dative subject and nominative object enter

17 For the sake of simplicity, I discuss the derivations of simple clauses here.18 Under the present perspective, T is raised to Fin after its Case and EPP features are de-leted by Agreeing with arguments.

28 Hideki Kishimoto

an Agree relation with T. Nevertheless, (54) suggests that even if the object is scrambled across the subject, the dative subject rather than the object undergoes A-movement to Spec-TP. The fact follows, given that scrambling is an adjunction operation, as often claimed (e.g. Saito 1986).

To be concrete, suppose that in (54), the object is first adjoined to vP (and then is moved to a higher position) via scrambling. When the object is moved and adjoined to vP, it occurs in a position structurally closer to T than the subject fill-ing Spec-vP, as illustrated in (55), (and the same holds true when it is moved by operator movement).

(55) [TP [vP OBJ [vP SUBJ V-v]] T ]

Nevertheless, the scrambled object in vP does not qualify as an argument to undergo A-movement to TP, because it appears in A’-position.19 If subject raising to TP at issue involves A-movement, the raising of the object to Spec-TP results in a non-uniform A-A’-A chain, which is illegitimate (Chomsky 1995). Thus, the subject rather than the scrambled object is raised to Spec-TP by A-movement, and the scrambled object is moved to a structural position higher than the subject fill-ing in Spec-TP in (54).

Under the present analysis, the possibility of subject raising is determined by the properties of tense. This analysis makes the further prediction that the raising of an oblique-marked subject to Spec-TP will be implemented if the clause has a nominative argument. This prediction is in fact borne out, for oblique subjects are susceptible to subject raising when they occur in a clause that includes a nomina-tive object.

(56) a. *Zutto kodomo-tati-de-sika hanasi-ga deki-nai-de i-ru. all.the.time child-pl-with-only talk-nom can.do-neg-ger be-pres ‘Only the children have been able to talk all the time.’ b. *Zutto hahaoya-kara-sika hanasi-ga deki-nai-de i-ru. all.the.time mother-from-only talk-nom can.do-neg-ger be-pres ‘Only the mother has been able to talk all the time.’

The examples in (56) differ from those in (50) in case marking. In (56), the object is marked with nominative case. (Note that dekiru, which can sanction nomina-tive case on its object, is a suppletive potential form of suru ‘do’.) In (56), the NPI object marked with nominative case is licensed by nai, but the NPI subject is not.20

19 It is sometimes claimed (e.g. Tada 1993) that clause-internal scrambling displays A-properties. At the same time, some researchers argue (e.g. Saito 1992) that clause-internal scrambling exhibits some A’-properties. I presume here that the mixed properties of scram-bling emerge from the fact that an argument moved by scrambling does not form an opera-tor-variable structure even though it appears in A’-position.20 One reviewer is not certain whether the examples in (56) are totally unacceptable. Al-though there might be some speaker variation in judgments here, the crucial point is that there exists a contrast in acceptability between (50) and (56).


The failure of the negative nai to license the oblique subject NPIs in the aspec-tual construction in (56) where the negative takes scope just over its complement clause gives us a good indication that oblique subjects are raised to the matrix Spec-TP when objects are marked with nominative case, as illustrated in (57a).

(57) a. [TP SUBJ-kara/de [vP SUBJ-kara/de OBJ-nom V-v]T] b. [TP [vP SUBJ-kara/de OBJ-acc V-v]T]

Note that subject raising is not implemented in the oblique-subject constructions when an object is marked with accusative rather than nominative case, as in (57b). These facts lead to the conclusion that subject raising is induced when tense bears the Case feature [+Nom] to value the Case feature of a nominative argument.

3.2. Naru-constructionsJapanese has another type of raising construction, which is constructed by embed-ding a clause under naru ‘become’ specifying the meaning of a change of state. The naru-construction shows syntactic behaviors different from those observed for the aspectual construction with the verb iru. The naru-construction takes a negated predicate or a clause introduced by yooni as its complement. It is shown that the subject of the main predicate, if subject raising applies, is extracted from the embedded clause in the naru-construction taking a yooni-complement, but not in the naru-construction which directly embeds a negated predicate.

First of all, observe that the predicate naru ‘become’ can take an adjectival predicate as its complement, as shown in (58).

(58) Ken-ga yasasiku nat-ta. Ken-nom kind become-past ‘Ken became kind.’

Since the verb naru can take an adjective as its complement, and since negative nai inflects like an adjective, a negated verbal clause can be embedded under the verb naru, as in (59).

(59) Ken-ga hanasa-naku nat-ta. Ken-nom talk-neg become-past ‘Ken stopped talking.’

Needless to say, the verb itself does not inflect like an adjective, so it is not possible for the verb to be embedded directly under naru.

Another type of complementation is possible with the predicate naru; the predicate naru can select a complement clause introduced by the complementizer yooni.

(60) Ken-ga hanasa-nai yooni nat-ta. Ken-nom talk-neg comp become-past ‘Ken stopped talking.’

In the variant of the naru-construction in (60), the embedded predicate does not

30 Hideki Kishimoto

have to be negated, as naru does not directly select it. Thus, the affirmative form hanasu ‘speak’ can appear in the complement clause, instead of the negated verb hanasa-nai ‘not speak’. Nevertheless, the complement clause is not finite, so the embedded verb cannot appear in the past form, i.e. the past form hanasi-ta ‘spoke’ cannot be placed in the complement clause.

In both variants of the naru-construction, represented by (59) and (60), inani-mate subjects and subject idioms are allowed to occur, as shown in (61).

(61) a. Kinoo-kara sora-ga {hare-naku/hare-nai yooni} nat-ta. yesterday-since sky-nom {clear-neg/clear-neg comp} become-past ‘The sky has not cleared since yesterday.’ b. Kono mise-de-wa kankodori-ga {naka-naku/naka-nai yooni} this store-at-top cuckoo-nom {sing-neg/sing-neg comp} nat-ta. become-past ‘Customers began to shop at this store.’

The data suggest that the upper verb does not impose a selectional restriction on the subject, and that the two types of naru-constructions possess a raising structure where the subject is generated in the complement clause.

The naru-constructions in (59) and (60) carry similar meanings, but the nomi-native subjects occupy different constituent positions: the subject of the predicate hanasu ‘speak’ in (59) remains in the embedded clause, as represented in (62a), but the subject of hanasu in (60) is raised to the matrix clause, as represented in (62b).

(62) a. [FinP [TP [vP [FinP [TP SUBJ [vP SUBJ OBJ V-v]T]T-Fin]nar]T]T-Fin] b. [FinP [TP SUBJ [vP [FinP [TP SUBJ [vP SUBJ OBJ V-v]T]T-Fin]nar]T]T-Fin]

In (59), no overt tense element appears in the complement clause selected by the predicate naru. Nevertheless, the embedded clause can be assumed to have a TP projection even though it is not visible in the overt strings, given Zanuttini’s (1997a) generalization that whenever NegP is projected in a clause, TP is also projected.

In both variants of the naru-construction given in (59) and (60), the structural position of subjects can be readily confirmed, since the negative nai appearing in the embedded clause does not take scope over the matrix TP. In the first type of naru-construction in (59), NPI subjects with sika are licensed irrespective of whether they are marked with nominative case or oblique case.

(63) a. Ken-sika hanasa-naku nat-ta. Ken-only talk-neg become-past ‘Everyone except Ken stopped talking.’ b. Ken-kara-sika kodomo-ni hanasi-o si-naku nat-ta. Ken-from-only child-to talk-acc do-neg become-past ‘Everyone except Ken stopped talking to the child.’


c. Kodomo-tati-de-sika atumara-naku nat-ta. child-pl-with-only gather-neg become-past ‘Everyone except the children stopped getting together.’

While all types of NPI subjects are licensed by the negative nai located in the embedded clause, a temporal adverb like kinoo-kara-sika ‘only from yesterday’ is not licensed.21(64) *Kinoo-kara-sika Ken-ga hanasa-naku nat-ta. yesterday-from-only Ken-nom talk-neg become-past ‘Ken stopped talking only from yesterday.’

The adverb kinoo-kara is allowed to appear in the construction only when the matrix predicate naru takes the past form, as in (65), which suggests that it is adjoined to the matrix TP.

(65) Kinoo-kara Ken-ga hanasa-naku {nat-ta/*nar-u}. yesterday-from Ken-nom talk-neg {become-past/become-pres} ‘Ken has stopped talking since yesterday.’

In (64), the temporal NPI is not licensed. This shows that the negative nai does not extend its scope over the matrix TP. On the other hand, in (63a), the nomi-native subject with sika is licensed by nai. This does not indicate that the subject remains in vP with no subject raising. Rather, the subject is raised to the embedded TP. The presence of subject raising in the embedded clause can be confirmed by the examples in (66).

(66) a. *Dare-ga gakusei-o home-mo {si-nakat-ta/si-naku anyone-nom student-acc praise-q {do-neg-past/do-neg nat-ta}. become-past} ‘Anyone {did not praise/stopped praising} the students.’ b. Sensei-ga dare-o home-mo {si-nakat-ta/si-naku teacher-nom anyone-acc praise-q {do-neg-past/do-neg nat-ta}. become-past} ‘The teacher {did not praise anyone/has stopped praising everyone}.’

As discussed in Kishimoto (2001), an indeterminate pronoun like dare ‘anyone’

21 If a temporal adverb is not anchored to a specific point of time, it allows the addition of the NPI sika, as in (i).

(i) Ken-ga yuusyoku-go-ni-sika hanasa-naku nat-ta. Ken-nom dinner-after-only talk-neg become-past ‘Ken stopped talking except after dinner.’

In (i), the NPI adverb yuusyoku-go-ni-sika ‘only after dinner’ is permitted, because it can be associated with the lower clause.

32 Hideki Kishimoto

is interpreted as an NPI when it is construed with mo. Note that a Q particle mo attached to the verb takes scope over vP. Then, the unacceptability of (66a), where mo fails to bind the indeterminate pronoun dare, suggests that the nominative sub-ject is moved out vP. On the other hand, the object is located inside vP, as shown by the acceptability of (66b). Importantly, the fact remains the same regardless of whether or not naru follows the verb, which shows that the nominative sub-ject is raised to the embedded TP, but not to the matrix TP in the first type of naru-construction.22

In the second type of naru-construction given in (60), naru takes a non-finite complement clause introduced by the complementizer yooni. This type of nai-con-struction, unlike the first type of naru-construction, does not allow a nominative subject to accompany sika.

(67) *Kaigi-de-wa Ken-sika hanasa-nai yooni nat-ta. meeting-at-top Ken-only talk-neg comp become-past ‘Everyone except Ken stopped talking at the meeting.’

Similarly, a temporal adjunct like kinoo-kara with sika is not licensed by the nega-tor nai in the lower clause.

(68) *Kinoo-kara-sika Ken-ga hanasa-nai yooni nat-ta. yesterday-from-only Ken-nom talk-neg comp become-past ‘Ken stopped talking only from yesterday.’

(68) shows that the temporal adjunct kinoo-kara-sika appears in the matrix clause. Then, the unacceptability of (67) suggests that the nominative subject appears in the matrix clause, which lies outside the scope of the negative nai. By contrast, oblique subjects with sika are licensed by nai, as in (69).

22 In contrast to nominative subjects, oblique subjects can be bound by mo attached to the verb, as shown in (i).

(i) a. Sono-go dare-kara kare-ni hanasikake-mo {si-nakat-ta/si-naku that-after anyone-from he-to talk-q {do-neg-past/do-ne nat-ta}. become-past} ‘{No one talked /Everyone stopped talking} to him after that.’ b. Dono hito-tati-de atumari-mo {si-nakat-ta/si-naku nat-ta}. any person-pl-with gather- q {do-neg-past/do-neg become-past} ‘{No one got/Everyone stopped getting} together.’

The examples suggest that the oblique subjects do not undergo subject raising even when they appear in the naru-constructions. One reviewer remarks that the sentences in (66) and (i) are awkward, though not unacceptable. The crucial point here is that the sentences comprised of obliquely-marked subjects are much better than those including nominative subjects.


(69) a. Ken-kara-sika kodomo-ni hanasi-o si-nai yooni nat-ta. Ken-from-only child-to talk-acc do-neg comp become-past ‘Everyone except Ken stopped talking to the children.’ b. Kodomo-tati-de-sika atumara-nai yooni nat-ta. child-pl-with-only gather-neg comp become-past ‘Everyone except the children stopped getting together.’

In the presence of an asymmetry in the licensing of subject NPIs with sika, it can be concluded that the nominative subject in (63a) undergoes subject raising to the matrix TP, while the oblique subjects in (63b-c) do not. (In (63b-c), no nomina-tive argument appears in the clause, so the oblique subjects remain in vP with no subject raising.)

In the second type of naru-construction in (60), the raising of the nomina-tive subject follows straightforwardly. In (60), the matrix T with the Case fea-ture [+Nom] licenses the nominative subject via Agree, while an EPP feature is assigned to the finite T in the matrix clause, and also to the lower non-finite TP (Chomsky 1995, Bošković 2002). (I postulate here that the two T-heads are assigned an EPP feature concomitantly, since no CP boundary intervenes between them.) Accordingly, when the subject is marked with nominative case, it is raised to the matrix TP, as in (70).

(70) [TP SUBJ [vP SUBJ [FinP [TP SUBJ [NegP [vP SUBJ OBJ V-v]Neg]T]Neg-T-Fin]nar]T]

In (70), the nominative subject falls outside the scope of negative nai, and hence, it cannot be turned into an NPI by the addition of sika.

In the first type of naru-construction in (59), just like the second type, its nom-inative subject is Case-licensed by the matrix T, and an EPP feature is assigned to T in the matrix clause, and also to the lower non-finite T (with no intervening CP). Note that the subject is not extracted from the embedded clause in this type of naru-construction. This fact falls out, given the assumption that the verb naru takes a null pronominal subject pro, as in (71).23(71) [TP pro [vP pro [FinP [TP SUBJ [NegP [vP SUBJ OBJ V-v]Neg]T]Neg-T-Fin]nar]T]

In (71), the lower T has an EPP feature, so the nominative subject is raised to the embedded TP. For the matrix T, its EPP requirement is met by pro. Consequently, the nominative subject of the main predicate remains in the embedded TP and the nominative subject NPI with sika is licensed in (63a).

The view that the nominative subjects occupy distinct constituent positions in the two types of naru-construction gains additional support from potential verb formation. To be concrete, in both variants of the naru-construction, it is possible to add a potential affix to the lower predicate.

23 Under the present perspective, naru is a verb taking an impersonal pronoun (like a weather predicate rain), and the structural difference between the two types of naru-con-structions is analogous, though not identical, to the difference obtained between John seems to be honest. and It seems that John is honest.

34 Hideki Kishimoto

(72) a. Ken-ga kanzyoo-o arawas-e-naku nat-ta. Ken-nom feeling-acc show-poten-neg become-past ‘Ken cannot show his feelings any longer.’ b. Ken-ga kanzyoo-o arawas-e-nai yooni nat-ta. Ken-nom feeling-acc show-poten-neg comp become-past ‘Ken cannot show his feelings any longer.’

In principle, potential verb formation is possible with verbs when they take sub-jects that refer to individuals having a control over the described events.24 The well-formedness of the sentences in (72) is naturally expected, since, in both cases, the lower predicate selects an animate subject. When the upper predicate is turned into a potential form, there arises a contrast in acceptability.

(73) a. *Ken-ga kanzyoo-o arawasa-naku nar-e-ta. Ken-nom feeling-acc show-neg become-poten-past ‘Ken can now keep a grip on his emotions.’ b. Ken-ga kanzyoo-o arawasa-nai yooni nar-e-ta. Ken-nom feeling-acc show-neg comp become-poten past ‘Ken can now keep a grip on his emotions.’

The predicate naru can be turned into the potential form if the embedded clause is introduced by yooni, as in (73b). This fact suggests that in the naru-construction taking a yooni-complement clause, the DP Ken is rendered as the subject of naru by moving through the matrix vP when it is raised to TP, as (70) illustrates. On the other hand, when naru takes a negated predicate as its complement, as in (73a), potential verb formation with naru fails. This fact suggests that the DP Ken does not occur in the matrix clause, as illustrated in (71).

Finally, it is worth noting that the subject of the embedded predicate is raised to Spec-TP if the embedded predicate is an adjective, as in (74) (=(58)).

(74) Ken-ga yasasiku nat-ta. Ken-nom kind become-past ‘Ken became kind.’

If the nominative subject is amenable to subject raising, the structure in (75) can be posited for the naru-construction in (74).

(75) [TP SUBJ [vP SUBJ [aP SUBJ isogasiku] nar]T]

In the type of construction where an adjective is directly embedded under naru, the matrix predicate naru can be turned into a potential form, as in (76).

24 The subjects of potential verbs need to refer to individuals having a control over a de-scribed event (potentially). Thus, (i) is not acceptable.

(i) *Ame-ga hur-e-ru. rain-nom fall-poten-pres ‘It can rain.’


(76) Mari-wa dare-ni-demo yasasiku nar-e-ru. Mari-top anyone-dat-even kind become-poten-pres ‘Mari can be kind to anyone.’

This fact suggests that the nominative subject is raised to TP via the vP projected from naru, so that it counts as the subject of the verb naru.25 Potential verb forma-tion provides empirical evidence that the subject resides in the matrix clause in this type of construction.

To summarize, this section has shown that in the naru-constructions where a negated verb is directly embedded under naru, subjects are moved only within the lower clause, even if subject raising applies. By contrast, in the naru-constructions where naru takes a yooni-complement, NPI subjects, if marked with nominative case, are not licensed by nai located in the embedded clause, showing that they are moved from the subordinate clause to the matrix clause.

4. ConclusionIn this article, on the basis of the syntactic behavior of NPIs with sika, it has been shown that the negative head nai functioning as a deadjectival negator undergoes head raising, while an adjectival negator retaining its categorical status as an adjec-tive does not. In adjectival clauses, nai retains its categorical status as an adjective, so Neg-head raising does not take place. Since the scope of negation does not extend over TP in clauses comprising an adjectival negator, a subject-object (or subject-complement) asymmetry is observed with regard to the licensing of NPIs (even in simple clauses).

Japanese invokes A-movement of subjects to Spec-TP when they are marked with either nominative or dative case. Subject raising is not implemented on oblique subjects when the clause does not include any nominative argument. In simple verbal clauses, the effects of subject raising are not detected by the behavior of NPIs, because a negative head undergoes head raising out of NegP to a struc-turally higher position where it takes scope over TP. Nevertheless, its effects can be detected in the raising construction where the negator is embedded under the aspectual verb iru ‘be’. In this type of raising construction, subjects appear outside the scope of the negative nai when they undergo subject raising. But when subjects remain in situ, they fall under the scope of negation. Thus, NPI subjects display distinct syntactic behaviors according to whether they undergo subject raising or not. Japanese has another type of raising construction (formed on the verb naru).

25 When an adjective is embedded under the verb naru ‘become’, the adjective can be turned into a subject-honorific form, as in (i).

(i) Sasaki-sensei-ga o-isogasiku nat-ta. Sasaki-teacher-nom hon-busy become-past ‘Ms. Sasaki became busy.’

This fact can also be taken to show that the subject located in the matrix clause is originated from the embedded clause.

36 Hideki Kishimoto

There are two variants of the naru-construction. In the naru-construction taking a yooni complement, the subject of the embedded verb is raised to the main clause when subject raising applies. In the naru-construction in which a negated verb is embedded under naru, the subject moves only within the embedded clause even when subject raising takes place.

Overall, in Japanese, the negative head nai undergoes Neg-head raising when it serves as a deadjectival negator, but not when it is an adjectival negator. The syn-tactic behavior of NPIs is affected by the two types of movement—subject raising and Neg-head raising.

ReferencesAoyagi, Hiroshi, and Toru Ishii (1994) On NPI licensing in Japanese. In: Noriko Akatsuka

(ed.) Japanese/Korean linguistics 4, 295–311. Stanford: CSLI Publications.Bošković, Željiko (2002) A-movement and the EPP. Syntax 5: 167–218.Carnie, Andrew (2006) Syntax: A generative introduction. Oxford: Blackwell.Chomsky, Noam (2000) Minimalist inquiries: The framework. In: Roger Martin, David

Michaels, and Juan Uriagereka (eds.) Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–155. Cambridge, MA: MIT Press.

Chomsky, Noam (2001) Derivation by phase. In: Michael Kenstowicz (ed.) Ken Hale: A life in language, 1–52. Cambridge, MA: MIT Press.

Chomsky, Noam (2004) Beyond explanatory adequacy. In: Adriana Belletti (ed.) Structures and beyond: The cartography of syntactic structures: Volume 3, 104–131. Oxford: Oxford University Press.

Chomsky, Noam (2008) On phases. In: Robert Freidin, Carlos P. Otero, and Maria Luisa Zubizarreta (eds.) Foundational issues in linguistic theory: Essays in honor of Jean-Roger Vergnaud, 133–166. Cambridge, MA: MIT Press.

Chomsky, Noam (2013) Problems of projection. Lingua 130: 33–49.Chomsky, Noam (2015) Problems of projection: Extensions. In: Elisa Di Domenico, Cor-

nelia Hamann and Simona Matteini (eds.) Structures, strategies and beyond: Studies in honor of Adriana Belletti, 3–16. Amsterdam: John Benjamins.

Fukui, Naoki (1995) Theory of projection in syntax. Stanford: CSLI Publications.Fukui, Naoki, and Hiromu Sakai (2003) The visibility guidance for functional categories:

Verb raising and related issues. Lingua 113: 321–375.Fukui, Naoki, and Yuji Takano (1998) Symmetry in syntax: Merge and demerge. Journal of

East Asian Linguistics 7: 27–86.Harada, S. I. (1976) Honorifics. In: Masayoshi Shibatani (ed.) Syntax and semantics 5: Japa-

nese generative grammar, 499–561. New York: Academic Press.Haegeman, Liliane (1995) The syntax of negation. Cambridge: Cambridge University Press.Hasegawa, Nobuko (2006) Honorifics. In: Martin Everaert and Henk van Riemsdijk (eds.)

The Blackwell companion to syntax, volume II, 493–543. Oxford: Blackwell.Inoue, Kazuko (1998) Sentences without nominative subjects in Japanese. Grant-in-aid for

COE research report (2A): Researching and verifying an advanced theory of human language, 1–34. Chiba: Kanda University of International Studies.

Inoue, Kazuko (2007) Case (with special reference to Japanese). In: Martin Everaert and Henk van Riemsdijk (eds.) The Blackwell companion to syntax, volume I, 295–373. Oxford: Blackwell.

Kato, Yasuhiko (1985) Negative sentences in Japanese. Sophia Linguistica 19. Tokyo: Sophia


University.Kato, Yasuhiko (1994) Negative polarity and movement. In: Masatoshi Koizumi and

Hiroyuki Ura (eds.) Formal approaches to Japanese linguistics 1: MIT working papers in Linguistics 24, 101–120. Cambridge, MA: MITWPL.

Kishimoto, Hideki (2001) Binding of indeterminate pronouns and clause structure in Japa-nese. Linguistic Inquiry 32: 597–633.

Kishimoto, Hideki (2005) Toogokoozoo-to bunpoo-kankei [Syntactic structures and grammatical relations]. Tokyo: Kurosio Publishers.

Kishimoto, Hideki (2007) Negative scope and head raising in Japanese. Lingua 117: 247–288.

Kishimoto, Hideki (2008) On the variability of negative scope in Japanese. Journal of Lin-guistics 44: 379–435.

Kishimoto, Hideki (2010) Subjects and constituent structure in Japanese. Linguistics 48: 629–670.

Kishimoto, Hideki (2012) Subject honorification and the position of subjects in Japanese. Journal of East Asian Linguistics 21: 1–41.

Kishimoto, Hideki (2013) Verbal complex formation and negation in Japanese. Lingua 135: 132–154.

Kishimoto, Hideki (2016a) Valency and case alternations in Japanese. In: Taro Kageyama and Wesley Jacobsen (eds.) Transitivity and valency alternations: Studies on Japanese and beyond, 125–154. Berlin: De Gruyter Mouton.

Kishimoto, Hideki (2016b) Bun-no koo-zoo to kaku-kankei [Sentence structures and case relations]. In: Murasugi, Keiko, Mamoru Saito, Yoichi Miyamoto, and Kensuke Takita (eds.) Nihongo bunpoo handobukku [Handbook of Japanese grammar], 102–145. Tokyo: Kaitakusha.

Kishimoto, Hideki (2017) Negative polarity, A-movement, and clause architecture in Japa-nese. Journal of East Asian Linguistics 17: 109–161.

Kitagawa, Yoshihisa (1986) Subjects in Japanese and English. Doctoral dissertation, Uni-versity of Massachusetts, Amherst.

Koizumi, Masatoshi (1999) Phrase structure in minimalist syntax. Tokyo: Hituzi Syobo.Koizumi, Masatoshi (2000) String vacuous overt verb raising. Journal of East Asian Linguis-

tics 9: 227–285.Koopman, Hilda, and Dominique Sportiche (1991) The position of subjects. Lingua 85:

211–258.Kuno, Susumu (1973) The structure of the Japanese language. Cambridge, MA: MIT Press.Kuroda, S.-Y. (1988) Whether we agree or not: A comparative syntax of English and Japa-

nese. Linguisticae Invesitgationes 12: 1–47.Miyagawa, Shigeru (1989a) Light verbs and the ergative hypothesis. Linguistic Inquiry 20:

659–668.Miyagawa, Shigeru (1989b) Syntax and semantics 22: Structure and case marking in Japanese.

San Diego: Academic Press.Miyagawa, Shigeru (2001) The EPP, scrambling and wh-in-situ. In: Michael Kenstowicz

(ed.) Ken Hale: A life in language, 293–338. Cambridge, MA: MIT Press.Miyagawa, Shigeru (2010) Why agree? Why move?: Unifying agreement-based and discourse-

configurational languages. Cambridge, MA: MIT Press.Otani, Kazuyo and John Whitman (1991) V-raising and VP-ellipsis. Linguistic Inquiry 22:

345–358.Pesetsky, David, and Esther Torrego (2001) T-to-C movement: Causes and consequences.

38 Hideki Kishimoto

In: Michael Kenstowicz (ed.) Ken Hale: A life in language, 355–426. Cambridge, MA: MIT Press.

Radford, Andrew (1997) Syntactic theory and the structure of English: A minimalist approach. Cambridge: Cambridge University Press.

Radford, Andrew (2009) Analysing English sentences: A minimalist approach. Cambridge: Cambridge University Press.

Rizzi, Luigi (1997) The fine structure of the left periphery. In: Haegeman, Liliane (ed.) Ele-ments of grammar: Handbook of generative syntax, 281–337. Dordrecht: Kluwer.

Rizzi, Luigi (2004) On the cartography of syntactic structures. In: Luigi Rizzi (ed.) The structure of CP and IP: The cartography of syntactic structures, volume 2, 3–15. Oxford: Oxford University Press.

Saito, Mamoru (1985) Some asymmetries in Japanese and their theoretical implications. Doctoral dissertation, MIT.

Saito, Mamoru (1992) Long distance scrambling in Japanese. Journal of East Asian Linguis-tics 1: 69–118.

Saito, Mamoru (2009) Optional A-scrambling. In: Yukinori Takubo, Tomohide Kinuhata, Szymon Grzelak and Kayo Nagai (eds.) Japanese/Korean linguistics 16, 44–63. Stanford, CA: CSLI Publications.

Sakai, Hiromu (1998) Feature checking and morphological merger. In: David J. Silva (ed.) Japanese/Korean linguistics 8, 189–201. Stanford, CA: CSLI Publications.

Shibata, Yoshiyuki (2014) Negative structure in Japanese. University of Pennsylvania Work-ing Papers in Linguistics 20: 291–299.

Shibata, Yoshiyuki (2015) Negative structure and object movement in Japanese. Journal of East Asian Linguistics 24: 217–269.

Shibatani, Masayoshi (1978) Nihongo-no bunseki [An analysis of Japanese]. Tokyo: Taishukan.Shimizu, Yasuyuki (2013) Gendaigo-ni-okeru kanzyookeiyoosi-no o-kaku-to gokousei

[Accusative case and word formation in Contemporary Japanese emotional adjectives]. Nihongo Bunpoo [Journal of Japanese Grammar] 13: 20–36.

Sportiche, Dominique (1988) A theory of floating quantifiers and its corollaries for con-stituent structure. Linguistic Inquiry 19: 429–449.

Tada, Hiroaki (1992) Nominative objects in Japanese. Journal of Japanese Linguistics 14: 91–108.

Tada, Hiroaki (1993) A/A-bar partition in derivation. Unpublished doctoral dissertation, MIT.

Takezawa, Koichi (1987) A configurational approach to case marking in Japanese. Unpub-lished doctoral dissertation, University of Washington.

Takezawa, Koichi (2004) Nihongo-hukugoo-zyutugo-ni-okeru hiteizi-no iti-to setu-koo-zoo [The position and structure of a negator in complex predicates in Japanese]. Nihongo bunpoo gakkai dai-go-kai happyoo ronbun shuu [Papers from the 5th meeting of the Society of Japanese Grammar], 175–184.

Takubo, Yukinori (2010) Nihongo-no koozoo [The structure of Japanese]. Tokyo: Kurosio Pub-lishers.

Ueda, Yukiko (2003) Subject positions in Japanese. In: Patricia M. Clancy (ed.) Japanese/Korean linguistics 11, 529–542. Stanford: CSLI Publications.

Ura, Hiroyuki (1999) Checking theory and grammatical functions in universal grammar. Oxford: Oxford University Press.

Watanabe, Akira (2004) The genesis of negative concord: Syntax and morphology of nega-tive doubling. Linguistic Inquiry 35: 559–612.


Zanuttini, Raffaella (1997a) Negation and verb movement. In: Liliane Haegeman (ed.) The new comparative syntax, 214–245. London: Longman.

Zanuttini, Raffaella (1997b) Negation and clause structure. New York: Oxford University Press.

Author’s contact information: [Received 11 January 2017; Graduate School of Humanities, Accepted 3 January 2018] Kobe University, 1-1 Rokkodai-cho, Nada-ku Kobe, Hyogo, 675-8501, Japan email: [email protected]

【要　旨】日本語の否定のスコープの投射

岸本　秀樹神戸大学

本稿では，文の否定辞のスコープの拡がり方を観察することにより，日本語において，主要部移動と名詞句移動の存在を確認することができることを論じる。日本語の否定辞は主要部移動を起こす要素で，移動が起こるかどうかによって，そのスコープの拡がり方が異なる。否定のスコープ内でのみ認可される否定極性表現の振る舞いから，日本語では，形容詞から脱範疇化により文法要素となった否定辞は主要部移動を起こし，形容詞の範疇的性質を残す否定辞は移動を起こさないこと，および，日本語の主語は，時制辞が主格の項を認可する場合に，文の主語位置への移動を起こすことを示す。また，「なる」に節が埋め込まれた複文では，主語の移動が起こった場合に，主節の主語位置に移動する構文と主語が埋め込み節内でのみ移動する構文があることも示す。

1 kishimoto - 日本言語学会Title 1_kishimoto.indd Created Date 4/27/2018 8:25:19 AM

Documents

Transcript of 1 kishimoto - 日本言語学会Title 1_kishimoto.indd Created Date 4/27/2018 8:25:19 AM